ggplot() is a plotting package based on the grammar of graphics. This is the most used plotting package, although some people still use baseplot (gasp!). Baseplot encompasses the built in plotting functionality in R. Don’t ask me anything about baseplot – I’ve NEVER used it! These days, ggplot is standard.

The current package version is ggplot2, but when calling ggplot, you will just use ggplot(). There is probably more documentation on ggplot compared to any other function or package in all of the R ecosystem! A few helpful resources:

ggplot cheatsheet R documentation for ggplot Carpentries ggplot lesson

There are also (free) books on this topic: Hadley Wickham’s book R Graphics Cookbook

A great document for more ggplot written by Sean Anderson

For when you become more advanced: ggplot extensions

Let’s get started.

ggplot() layers upon it self, so the first command is always ggplot(). Then, you add a geom(). There are SO MANY geoms. The most common are: geom_point(), geom_line(), geom_histogram(), geom_ribbon(), geom_boxplot(), geom_pointrange(). Let’s try a few.

# geom histogram -- let's build a histogram of hindfoot lengths

ggplot() +
    geom_histogram(data = dat, aes(hindfoot_length))
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).

# we get some warnings -- first, pick better binwidths and second, 3348 rows were removed.
# Why?

# you can put the data and mapping (aes -- aesthetics) 
#in the ggplot() call and get the same plot

ggplot(data = dat, aes(hindfoot_length)) +
    geom_histogram()
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).

# the difference is that anything in the () associated with 'ggplot' 
# are now global -- they will relate to anything else you add

# in contrast, if you put the data in the geom specifically,
# it will only apply to that geom

# this is useful for when you want to add additional 
# datasets to the same plot! we will do this later.

# you can save your plot to an object

hindft_hist <- ggplot(data = dat, aes(hindfoot_length)) +
    geom_histogram()

# nothing happens when you run the above! 
# to see the plot, call the object

hindft_hist
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).

# we will learn more about themes and other customizations in the last lesson
# for right now, use theme_classic()

hindft_hist <- ggplot(data = dat, aes(hindfoot_length)) +
    geom_histogram() +
    theme_classic()

hindft_hist
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).

# what changed?

# save your plot to your working directory

ggsave(hindft_hist, file = "hindft_hist.png",
             height = 10, width = 10)
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).
# you can also integrate the %>% with ggplot 
# (I don't do this very often because I'm usually saving 
# plots to objects so I can save them to my local machine)

dat %>%
    ggplot(aes(hindfoot_length)) +
    geom_histogram() +
    theme_classic()
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## Warning: Removed 3348 rows containing non-finite values (stat_bin).

Let’s build a scatterplot.

ggplot() +
    geom_point(data = dat, aes(x = weight, y = hindfoot_length))
## Warning: Removed 4048 rows containing missing values (geom_point).

# what if we wanted to color this by species?
ggplot() +
    geom_point(data = dat, aes(x = weight, y = hindfoot_length, color = species_name))
## Warning: Removed 4048 rows containing missing values (geom_point).

# later, we will see how to assess whether there is a relationship between
# hindfoot_length and weight and how to plot that relationship

# what if you wanted to plot the logged hindfoot data?
ggplot() +
    geom_point(data = dat, aes(x = weight, y = log_hindfoot_length, color = species_name))
## Warning: Removed 4048 rows containing missing values (geom_point).

# or,
ggplot() +
    geom_point(data = dat, aes(x = weight, y =log10(hindfoot_length), color = species_name))
## Warning: Removed 4048 rows containing missing values (geom_point).

What about a time series? Do the number of plots per year change?

# we first have to do a bit of summarizing
plot_sum <- dat %>%
    group_by(plot_id, year) %>%
    summarize(n()) %>%
    rename(count = 'n()')
## `summarise()` has grouped output by 'plot_id'. You can override using the
## `.groups` argument.
ggplot(plot_sum) +
    geom_line(aes(x = year, y = count, group = plot_id))

# we just used 'group', not 'color'
ggplot(plot_sum) +
    geom_line(aes(x = year, y = count, group = plot_id, color = plot_id))

# we can change the colors (we will get more into this later)
length(unique(dat$plot_id)) # how many unique plot types are there
## [1] 24
# let's use Rcolorbrewer]
#(https://cran.r-project.org/web/packages/RColorBrewer/RColorBrewer.pdf) to get new colors

#install.packages("Rcolorbrewer")
#library(RColorBrewer) # remember you have to install and load packages

# to get individual color hex codes, color-hex.com is FAB!

colfunc <- colorRampPalette(c("#00c913", "#000000")) 
col_list <- colfunc(24)
# this will give us 24 colors between green and black
# will look yucky, but just for show

# will have to create a factor of plot_id (one useful purpose of factors)
dat <- dat %>%
    mutate(plot_id_f = as.factor(plot_id))

plot_sum <- dat %>%
    group_by(plot_id, plot_id_f, year) %>%
    summarize(n()) %>%
    rename(count = 'n()')
## `summarise()` has grouped output by 'plot_id', 'plot_id_f'. You can override
## using the `.groups` argument.
ggplot(plot_sum) +
    geom_line(aes(x = year, y = count, group = plot_id, color = plot_id_f)) +
    scale_color_manual(values = col_list)

# Why do you think we had to turn plot_id into a factor?

a good guide to different ways to add color – we will look at some of these later. https://stackoverflow.com/questions/70942728/understanding-color-scales-in-ggplot2

Note this link takes you to stackoverflow. 99.9% of the time, you can find the answer to your coding issue there. I probably look up something on stackoverflow everyday!

Let’s look at some other types of plots.

# boxplot
ggplot(dat) +
    geom_boxplot(aes(x = species_name, y = hindfoot_length))
## Warning: Removed 3348 rows containing non-finite values (stat_boxplot).

Challenge: How would you add color to this?

ggplot(dat) +
    geom_boxplot(aes(x = species_name, y = hindfoot_length, color = species_name))
## Warning: Removed 3348 rows containing non-finite values (stat_boxplot).

Challenge: What if you want to remove the species with no data? How would you do this?

dat_no_NA <- na.omit(dat)

ggplot(dat_no_NA) +
    geom_boxplot(aes(x = species_name, y = hindfoot_length, color = species_name))

Let’s add points to our boxplot. Then we can see why it is useful to pay attention to the positioning of the dataset and the geoms.

ggplot(dat_no_NA) +
    geom_boxplot(aes(x = species_name, y = hindfoot_length, color = species_name)) +
    geom_point()
## Error in `check_required_aesthetics()`:
## ! geom_point requires the following missing aesthetics: x and y
# why do we get an error?

ggplot(dat_no_NA, 
             aes(x = species_name, y = hindfoot_length, color = species_name)) +
    geom_boxplot() +
    geom_point()

# why did we not get an error?

# let's make some of the points translucent
ggplot(dat_no_NA, 
             aes(x = species_name, y = hindfoot_length, color = species_name)) +
    geom_boxplot() +
    geom_point(alpha = 0.2)

# what if we wanted the points in the background? switch the order! ggplot is iterative
ggplot(dat_no_NA, 
             aes(x = species_name, y = hindfoot_length, color = species_name)) +
    geom_point(alpha = 0.2) +
    geom_boxplot() 

# we can 'jitter' the points so they aren't all on top of each other
ggplot(dat_no_NA, 
             aes(x = species_name, y = hindfoot_length, color = species_name)) +
    geom_jitter(alpha = 0.2) +
    geom_boxplot()

Let’s look at pointrange. This is helpful when you want to plot error around a data point. We will have to do some data wrangling to get our data in the format we want for this plot.

point_error_dat <- dat %>%
    na.omit() %>%
    group_by(species_name) %>%
    summarise(mean_hfl = mean(hindfoot_length),
                        sd_hfl = sd(hindfoot_length),
                        lwr = mean_hfl - sd_hfl,
                        upr = mean_hfl + sd_hfl)

# plot the mean + and - 1 SD
ggplot(point_error_dat) +
    geom_pointrange(aes(x = species_name,
                                            y = mean_hfl,
                                            ymin = lwr,
                                            ymax = upr))

One cool thing in ggplot is faceting. You can easily make several plots by group.

ggplot(dat_no_NA, aes(x = weight, y = hindfoot_length)) +
    geom_point() +
    facet_wrap(~ species_name)

# there is also facet_grid, which is helpful for defining rows and columns
ggplot(dat_no_NA, aes(x = weight, y = hindfoot_length)) +
    geom_point() +
    facet_grid(sex ~ species_name)

# what do you think facet_wrap does given you see what facet_grid does?

# these are fully customizable and we will go over that later

That is it for now. Let’s move into data analysis!